Nonlinear Multiple Regression Models

Abstract

In this chapter, nonlinear transformations are introduced that expand linear regression options to include situations in which marginal responses are either increasing or decreasing, rather than constant. We will explore Tukey’s Ladder of Powers to identify particular ways to rescale variables to produce valid models with superior fit. An example will be offered in the context of naïve models built for forecasting, and in Chapter 14, examples with explanatory multiple regression models will be added.